{"id":"a563d2f6-a8e9-419a-932c-fc863d4debef","arxiv_id":"2506.23002","paper_version":1,"verdict":"REJECT","confidence":"MODERATE","novelty_score":3.0,"correctness_risk":"high","formal_verification":"none","parameter_count":1,"one_line_summary":"A simple image processing pipeline (grayscale, contrast enhancement, binarization) is shown to reduce blur-related errors in smartphone-to-smartphone visible light communication.","lead":"This paper applies standard image processing steps (grayscale conversion, contrast stretching, and adaptive thresholding) to reduce blur in smartphone-to-smartphone visible light communication images. The technique is reported to improve data recovery to 96% under varied distance, tilt, rotation, and lighting conditions.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 96% recovery-efficiency claim rests on an undocumented bit-to-cell mapping; without synchronization details the BER curves and the headline improvement are not reproducible.","rationale":"The reader identified the missing synchronization and bit-to-cell mapping as the weakest assumption; I agree and treat it as the load-bearing concern because every BER number and the 96% claim inherits from that mapping. The paper gives no code, no raw dataset, no camera exposure settings, no number of repeated trials, and no explicit definition of recovery efficiency. While the algorithm itself is simple and standard, the advertised quantitative result cannot be verified without knowing how photographed pixels are aligned to the transmitted cells. I also note an internal tension: the quoted BER values at tilt and rotation extremes (≈0.385 and ≈0.35) imply much lower recovery efficiency than 96% if efficiency is 1−BER, so the headline is ambiguous at best. These reproducibility and consistency problems support the reader's reject verdict, so no verdict change is needed.","tokens_in":9353,"tokens_out":3517,"duration_ms":40986,"concrete_test":"Run the described pipeline on one raw captured frame from each experiment (distance, tilt, rotation) with the transmitted 200×200 bit pattern known. Implement the bit-to-cell mapping two ways: (a) using a manually annotated screen-corner/grid overlay as ground-truth alignment, and (b) using only the Section V pipeline with an automated threshold and no external alignment information. If the BER under the automated mapping is not reported, or if it matches the paper's BER only when the manual overlay is used, the 96% claim is not supported. Also compute recovery efficiency as 1−BER for every operating point and check whether any condition reaches 0.96.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim (abstract and Sections VI–VII: the algorithm 'improved ... by 96%') depends entirely on BER values computed from captured binarized images. The pipeline in Section V converts the Rx image to grayscale, applies contrast enhancement, scaling, and adaptive thresholding, but it never describes how the resulting binary image is mapped back to the transmitted 40,000-bit sequence. A 200×200 encoded frame becomes a photographed grid whose cell positions shift with distance, tilt, and rotation; without a documented grid-detection, perspective-correction, or frame-synchronization step, the correspondence between image pixels and transmitted bits is indeterminate. Furthermore, Section VI reports final BER values around 0.385 at tilt extremes and 0.35 at rotation extremes; if recovery efficiency is 1−BER, those numbers imply roughly 61.5% and 65% efficiency, not 96%. No baseline BER or explicit definition connects these quoted values to the 96% figure. The headline improvement is therefore neither reproducible from the described method nor internally consistent with the paper's own numerical results.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a low-complexity blur reduction technique for smartphone-to-smartphone visible light communication (S2SVLC). The receiver-side algorithm converts a captured RGB image to grayscale, applies contrast enhancement, scaling, and adaptive thresholding, and claims this reduces bit errors compared to a conventional system without image processing. Experiments vary link distance (0–50 cm), tilt angle (−50° to 50°), rotation angle (−40° to 40° in Section VI, though Table I says −50° to 50°), and illumination (ambient vs. dark), transmitting 40,000 bits encoded as a 200×200 cell image. The abstract and conclusion state that the technique 'improve[s] the recovery efficiency to 96% at the receiver end at different conditions.' The central claim is that the algorithm substantially reduces BER across these conditions, but the paper provides no definition of the 96% metric, no synchronization or bit-to-cell mapping procedure, and no statistical characterization of the BER measurements.","tokens_in":9529,"tokens_out":4622,"duration_ms":52058,"significance":"If the 96% recovery-efficiency claim were properly defined and supported, the work would offer a practically attractive, low-complexity pre-processing step for smartphone-based optical camera communication, a topic of ongoing interest in the VLC community. The paper presents a real measurement campaign over distance, tilt, rotation, and lighting, which is appropriate for a systems paper. However, the contribution as written is not reproducible: the headline metric is undefined and internally inconsistent with the reported BER values, the BER computation pipeline lacks a documented synchronization/alignment stage, and the comparison baseline is not quantified. The paper also relies heavily on the authors' own prior publications for the core algorithm steps without clearly delineating the new contribution. These issues are load-bearing for the central claim, and I cannot recommend acceptance or even major revision without substantial rework.","major_comments":[{"comment":"The paper's central claim—that the proposed technique improves 'recovery efficiency to 96%'—is never defined. No equation or text in Sections V–VII defines what 'recovery efficiency' means (e.g., 1−BER, fraction of recovered bits, or fraction of frames decoded). The reported BER values at tilt extremes and rotation extremes are approximately 0.385 (Section VI, Fig. 7) and 0.35 (Section VI, Fig. 9), respectively; if efficiency is 1−BER, these correspond to 61.5% and 65%, not 96%. No baseline BER is reported for the 'conventional system' against which the 96% improvement is computed. The headline claim is therefore neither definable nor derivable from the data presented.","section":"Abstract; Section VI, Figs. 5, 7, 9; Conclusion"},{"comment":"The paper never describes how the binarized image is mapped back to the transmitted 40,000-bit sequence. The receiver captures a photograph of a 200×200 cell frame; with varying distance, tilt, and rotation, the cell grid undergoes perspective distortion. The algorithm description in Section V stops at adaptive thresholding and provides no grid detection, perspective correction, or frame synchronization step. Without a documented alignment procedure, the BER values in Figures 5, 7, and 9 cannot be reproduced, and may measure misregistration rather than decoding errors. This is a fundamental reproducibility gap for the paper's only quantitative performance evidence.","section":"Section V (Steps 1–4); Table I; Section VI"},{"comment":"The BER results are reported without any statistical support: no error bars, no number of repeated trials, and no confidence intervals. The text also contains contradictory phrasing: for tilt and rotation, it states that 'the BER increases with the increasing tilt/rotation angles' but then says 'The bit error rate reduces to around 0.385' and 'reduces to around 0.35.' The intended meaning is presumably that the BER rises to about 0.385/0.35 at the extreme angles, but the wording is ambiguous. In addition, no quantitative BER values for the 'conventional system' are given in the figures or text, so the claimed improvement over baseline cannot be verified.","section":"Section VI; Figures 5, 7, 9"},{"comment":"The novelty boundary of the proposed algorithm relative to the authors' own prior work is not articulated. Section II introduces the proposed method as 'novel,' but the algorithm steps in Section V cite the authors' previous publications [4] and [5] ('Data Detection Technique' and 'Non-Blind Image Restoration Technique') without specifying what is new in this manuscript. Since the paper's contribution rests on the algorithm, the reader cannot determine whether this is an incremental extension of prior work or a genuinely new technique. The manuscript should clearly separate the new contribution from the previously published components.","section":"Section II; References [4]–[7]"}],"minor_comments":[{"comment":"The sentence 'The key technique it to avoid the repeated scanning of the transmitted data' contains a grammatical error ('it to' should be 'is to').","section":"Abstract"},{"comment":"Equation numbering is inconsistent: two equations are labeled (4), and the text refers to 'In (4)' for the contrast-stretch formula while Equation (2) and (3) appear earlier. The variables a, b, c, d in the contrast-stretch equation are introduced but their relationship to the text is confusing.","section":"Section IV"},{"comment":"Several figure references do not match the captions: Section V.A refers to 'Fig. 2' for the received blurred image but the actual system block diagram is Figure 2; Section VI first paragraph refers to 'Fig 4' for the real-time setup, but the real-time setup is Figure 3; and the schematic in Figure 4 is labeled 'Fig. 4' in the text as BER vs distance. Please renumber figures or correct in-text references.","section":"Figure numbering"},{"comment":"The rotation range is inconsistent: Table I states rotation angles of '-50 t0 50 degrees' (with a typo 't0'), while the rotation experiment in Section VI says '-40 to 40 degrees' and later mentions '(-50,50) degrees'. Clarify the actual experimental range.","section":"Section VI; Table I"},{"comment":"The reference list has multiple numbering errors: [5] is listed as a non-blind restoration paper but is cited as TETRIS in the text; [6] is cited as the flashing/surfing method but the reference list entry is about hemoglobin estimation; [13] is used for both SoftLight and a color-spaces paper; [22] is duplicated. These errors impede the reader's ability to trace the literature.","section":"References"},{"comment":"The description of the 'no light' condition is incomplete: only the ambient case is quantified (123 lumens), and the dark-room condition (e.g., residual screen illumination, any standby lighting) is not specified, which affects reproducibility.","section":"Section VI"}],"recommendation":"reject","confidential_remarks":"The manuscript's reference list and attribution patterns warrant editorial scrutiny: the core algorithm steps are largely drawn from the authors' own prior conference papers [4] and [5], and the relation between this submission and those works is not stated. The reference numbering is also badly scrambled, which may indicate a carelessly assembled manuscript. For the journal's scope, an experimental OCC paper with an undefined headline metric and no synchronization description does not provide a usable contribution; even with major revision, the paper would require new experiments or a fundamentally clearer account of the measurement pipeline to support the claimed 96% figure."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Punchline: the experimental effort is real, but the headline number doesn't survive contact with the paper's own text. The 96% recovery-efficiency claim is undefined, the BER figures at tilt/rotation extremes imply much lower recovery, and the bit-to-cell mapping is never described, so the BER curves are not reproducible. The core algorithm was also published by the same authors in a 2022 CSNDSP paper (ref [7]).\n\nWhat's genuinely new: a set of parameter sweeps — distance 0–50 cm, tilt ±50°, rotation ±50°, ambient vs dark — for a screen-to-camera OCC link using their preprocessing chain. That chain (RGB→grayscale, contrast stretch, scaling, adaptive thresholding) is textbook image processing, so the contribution is the systematic measurement, not the method. I give them credit for testing both ASCII and QR coding and for working in realistic conditions.\n\nThe soft spots are load-bearing. The abstract and conclusion repeat the 96% figure, but Section VI gives no definition. Is it recovered bits divided by transmitted bits? A relative improvement over the unprocessed case? The text says 'the bit error rate reduces to around 0.385' at the tilt extremes and 0.35 at the rotation extremes. A BER of 0.385 means roughly 61.5% of bits are correct, which is a long way from 96%. Without a baseline or a formula, the headline claim is empty. Second, there is no documented grid-detection, perspective-correction, or frame-synchronization step. The receiver photographs a 200×200 cell grid; with tilt and rotation the cell positions shift. How the binary image is aligned with the transmitted 40,000-bit sequence is never explained, so the BER values can't be independently checked. Third, there are no error bars, no repeated trials, and no statistical analysis. For a measurement paper that's a basic omission. The prose is also rough: duplicated lines, duplicate reference [13], and phrases like 'the bit error rate reduces to around 0.385' that are simply confusing.\n\nThe central argument is not hopeless. It is plausible that simple preprocessing reduces blur-induced errors in phone-to-phone links, and the qualitative trends (BER rises with distance, tilt, rotation) are sensible. But the evidence as presented does not support the 96% number. This is a desk-reject level problem: the main claim is not defined or reproducible.\n\nWho this is for: researchers already inside the smartphone OCC subfield who want a quick look at qualitative trends. I wouldn't cite it, and I wouldn't spend referee time on it as is. If the authors define the metric, add a proper synchronization description, and report error bars, a revised version could be worthwhile.","headline":"A systematic but under-specified experimental report whose headline 96% recovery-efficiency claim is not defined or reproducible; the algorithm itself is not new.","tokens_in":10072,"tokens_out":3887,"would_cite":false,"duration_ms":39216,"reading_group":"no","serious_thinker":"yes","would_accept_peer_review":false},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proposes a receiver-side blur-reduction pipeline for smartphone-to-smartphone visible light communication and reports that it improves recovery efficiency to 96% across varied distance, tilt, rotation, and lighting conditions.","keywords":["visible light communication","optical camera communication","smartphone-to-smartphone","image blur reduction","grayscale conversion","adaptive thresholding","OOK","QR code"],"falsifier":"Transmit a known 40,000-bit sequence, apply the four-step pipeline, and compare the decoded bits to the original using an independent synchronization method that does not use the decoded payload to find the alignment; if the 96% recovery efficiency drops substantially under this check, the central claim fails.","tokens_in":9160,"feed_emoji":"📱","tokens_out":6453,"duration_ms":65202,"temperature":0.7,"pith_summary":"The paper is trying to establish that a simple, low-complexity image processing chain—grayscale conversion, contrast enhancement, scaling, and adaptive binarization—can rescue blurred frames in smartphone-to-smartphone visible light communication (S2SVLC) instead of discarding them. If true, the system recovers 96% of transmitted bits at the receiver across distances up to 50 cm, tilt and rotation spans, and both ambient and dark surroundings, while also supporting a higher data rate. The motivation is practical: S2SVLC links based on screen-to-camera optical communication suffer blur from distance, misalignment, and lighting, and existing restoration methods are either too slow or need noise statistics unavailable at the receiver. The paper tests the pipeline with ASCII and QR coding using a real phone-to-phone setup and reports lower bit-error rates than the conventional no-image-processing system.","feed_headline":"Blur-cleaning pipeline lifts phone-to-phone VLC recovery to 96%","feed_subtitle":"A four-step image processing chain recovers blurred screen-camera frames across distance, tilt, rotation, and lighting conditions.","key_machinery":"The load-bearing mechanism is the blur reduction algorithm's own pipeline, implemented at the receiver. Step 1 maps RGB to grayscale, either by equal average or weighted luminance, reducing leakage between color channels. Step 2 applies contrast stretching so dark and bright cells separate. Step 3 normalizes minimum and maximum pixel values to set a detection threshold. Step 4 uses adaptive thresholding, which chooses per-region thresholds, to binarize the image into white and black cells that are then converted back into OOK bits. The central decision is a thresholding decision: each cell's value after scaling is compared to an adaptively chosen level, turning a blurred analog image into a clean binary representation without knowing the noise variance.","core_discovery":"On its own terms, the paper's central claim is that blur in S2SVLC can be largely corrected by a four-step receiver-side algorithm: convert the captured RGB frame to a single-channel grayscale image, apply contrast stretching, normalize the gray-level range, and binarize with adaptive thresholding. Because the transmitter encodes each bit as a white or black cell in an MxN frame under OOK-NRZ, the binarized output can be read as a bit stream. The paper reports that this processing improves recovery efficiency to about 96% at the receiver, lowers bit-error rate in ambient and no-light conditions over a 0–50 cm link span, and keeps bit-error rate below the conventional system across tilt and rotation experiments. The paper also argues that the method is less complex than HSV-based systems and non-blind or blind restoration, because it avoids repeated scanning and does not need noise statistics or high-computation networks. The claim covers both ASCII and QR encoded data.","pith_inferences":["The 96% recovery-efficiency figure is only as strong as the unstated frame-synchronization and bit-to-cell alignment step; a reader should treat it as an upper bound on decoding accuracy until that step is specified.","A natural testable extension is to apply the same four-step pipeline to LED-to-camera links, where rolling-shutter blur and cell-boundary smearing are analogous but the transmitter is not a screen.","One could also benchmark the pipeline against learned deblurring on the same captured frames; the paper's complexity argument would be strengthened if the simple pipeline matched or beat learned methods on bit-error rate per frame."],"forward_implications":["The proposed pipeline can be applied to standard QR and ASCII encoded frames, so it does not require customized barcodes like earlier systems.","Bit-error rate decreases in both ambient-light and dark-room conditions compared with the conventional receiver, and the gain persists across the tested 0–50 cm distance range.","Because blurred frames are processed rather than discarded, fewer retransmissions or rescans are needed, which the paper says raises the achievable data rate.","Recovery efficiency reaches about 96%, meaning only a small fraction of the transmitted bits is lost at the receiver across the tested conditions.","The method's low complexity makes it more feasible for real-time smartphone processing than HSV-based or neural restoration approaches."],"supporting_citations":[{"why":"Defines the screen-to-camera data detection problem and the cell-based OOK frame structure the proposed algorithm decodes.","marker":"[4]"},{"why":"Prior screen-to-camera system whose color-block blur limitation motivates the binarization-based blur-aware receiver.","marker":"[5]"},{"why":"Earlier version of this blur reduction method for QR and ASCII coding, which the paper extends with the full four-step pipeline and bit-error-rate measurements.","marker":"[7]"},{"why":"Prior blur equalization approach for screen-to-camera optical camera communication that the proposed method is positioned against.","marker":"[10]"},{"why":"Groundwork screen-to-smartphone system whose color-space processing latency motivates the faster grayscale route.","marker":"[11]"},{"why":"Identifies dynamic environments, lens distortion, angle, and distance as causes of blur, which are exactly the conditions tested here.","marker":"[12]"},{"why":"Prior screen-camera link using a different color space whose false-negative soft hint problem supports the grayscale modulation choice.","marker":"[13]"},{"why":"Provides the contrast enhancement operation used in Step 2 of the proposed algorithm.","marker":"[28]"},{"why":"Supplies the contrast stretching formula used to scale pixel values.","marker":"[29]"}],"fun_headline_variants":["Phone-to-phone VLC blur fix hits 96% recovery","Four-step image fix sharpens VLC data recovery to 96%","Blur removal lifts phone VLC link recovery to 96%","Image processing boosts VLC recovery to 96% in smartphone links","96% recovery achieved with blur-reducing image pipeline for VLC"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The numbers assume the receiver can reliably synchronize and align each captured frame with the transmitted 40,000-bit sequence before errors are counted; the paper does not describe that alignment procedure, so the reported error rates and the 96% figure depend on an unstated synchronization step.","fun_headline_variants_meta":{"raw":{"variants":["Phone-to-phone VLC blur fix hits 96% recovery","Four-step image fix sharpens VLC data recovery to 96%","Blur removal lifts phone VLC link recovery to 96%","Image processing boosts VLC recovery to 96% in smartphone links","96% recovery achieved with blur-reducing image pipeline for VLC"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000806,"raw_usage":{"total_tokens":3541,"prompt_tokens":950,"completion_tokens":2591,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":566,"completion_tokens_details":{"reasoning_tokens":2499}},"tokens_in":566,"tokens_out":2591,"duration_ms":18527,"temperature":1.0,"reasoning_tokens":2499,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T21:51:46.990339+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Transmit a known 40,000-bit sequence, apply the four-step pipeline, and compare the decoded bits to the original using an independent synchronization method that does not use the decoded payload to find the alignment; if the 96% recovery efficiency drops substantially under this check, the central claim fails.","supporting_citations":[{"cited_title":"Data Detection Technique for Screen -to-Camera Based Optical Camera Communications,","cited_arxiv_id":null,"evidence_quote":"Defines the screen-to-camera data detection problem and the cell-based OOK frame structure the proposed algorithm decodes."},{"cited_title":"Non-Blind Image Restoration Technique in Screen –to– Camera based Optical Camera Communications,","cited_arxiv_id":null,"evidence_quote":"Prior screen-to-camera system whose color-block blur limitation motivates the binarization-based blur-aware receiver."},{"cited_title":"QR Code Scanning app for Mobile Devices","cited_arxiv_id":null,"evidence_quote":"Prior blur equalization approach for screen-to-camera optical camera communication that the proposed method is positioned against."},{"cited_title":"A Blur Equalization Method for Screen-to-Camera Based Optical Camera Communications,","cited_arxiv_id":null,"evidence_quote":"Groundwork screen-to-smartphone system whose color-space processing latency motivates the faster grayscale route."},{"cited_title":"COBRA:Color Barcode Streaming for Smartphone System","cited_arxiv_id":null,"evidence_quote":"Identifies dynamic environments, lens distortion, angle, and distance as causes of blur, which are exactly the conditions tested here."},{"cited_title":"Rain Bar: Robust Application-Driven Visual Communication Using Color Barcodes","cited_arxiv_id":null,"evidence_quote":"Prior screen-camera link using a different color space whose false-negative soft hint problem supports the grayscale modulation choice."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the contrast enhancement operation used in Step 2 of the proposed algorithm."},{"cited_title":"PLOS ONE7(1):e29740","cited_arxiv_id":null,"evidence_quote":"Supplies the contrast stretching formula used to scale pixel values."}],"review_version":1}